On 10/15/05, Cyril Pernet <[log in to unmask]> wrote:

Hello Maria

I think I understand what they did .. I said I think .. note so sure
They set a model were they entered as regressors the 6 motions
parameters, and extra parameters directly extracted from the data (whole
brain signal from a region ???, signal from ventricle and signal from
white matter). So now you have a model of regression with 10 regressors
(the 9 above + cste). Note that it sounds strange to me to perform a
global normalization during the preprocessing and to include the whole
brain signal from a region (but what it is?) and a cste reg that is the
mean over all voxels .. (whole brain sig = cste ?? or whole brain sig
already scale by the global normalization - maybe someone has an idea
?). Anyway, from that, I guess, they obtained a new time series where
all this stuff is regressed out.. i.e. they got the residuals and then
performed the correlation analysis on residuals between ROIs. Thus your
model should incorporate this kind of regressors .. but the goal is to
clean the data and thus you don't need 2 conditions.

Good luck
Cyril